1. Field of the Invention
The present invention relates to an image processing device, an image processing method, and a computer-readable medium by which an augmented reality marker within a captured image is recognized.
2. Description of the Related Art
Mobile terminal devices, such as smartphones, are becoming increasingly multifunctional, and generally have an image processing function referred to as Augmented Reality (AR), in addition to a camera function (imaging function). In this function, a virtual image (such as a character image) is displayed in a predetermined area within a captured image by image composition. Specifically, when a photographic subject is captured by the imaging function with an AR marker composed of printed material (a marker on which a white pattern is drawn within a square black frame) being arranged at or near the position of the subject, the mobile terminal device recognizes the AR marker included in the captured image by analyzing the captured image, and displays a virtual image on the area of the AR marker by image composition. In this processing, in order to recognize the AR marker, the mobile terminal device converts the captured image to, for example, a gray scale image, and binarizes it using an adaptive thresholding. This technology to binarize an image using an adaptive thresholding is described in, for example, Japanese Patent Application Laid-open (Kokai) Publication No. 07-057080.
In this technology to binarize an image using adaptive thresholding, binarization processing on a captured image is performed for each frame. However, this processing is too CPU-intensive for mobile terminal devices such as smartphones. Therefore, there is a problem in that processing speed becomes slow and AR markers cannot be smoothly recognized.
This problem may be solved by binarizing captured images by a fixed threshold binarization method using a fixed binarization threshold. However, in this configuration, because a binarization threshold for the entire captured image is calculated, the calculation of an optimal binarization threshold within an AR marker becomes difficult. Accordingly, there is a problem in that, although processing speed is high, recognition rate is low.